Determination of the Effects of Transcutaneous Auricular Vagus Nerve Stimulation on the Heart Rate Variability Using a Machine Learning Pipeline.

Abstract

Background:We are all aware of day-to-day healthy stress, but, when sustained for long periods, stress is believed to lead to serious physical and mental health issues.Materials and Methods:In this study, we investigated the potential effects of transcutaneous auricular vagus nerve stimulation (taVNS) on stress processing as reflected in the electrocardiogram (ECG)-derived biomarkers of stress adaptability. Stress reflecting biomarkers included a range of heart rate variability metrics: standard deviation of N-N intervals (SDNN), root mean squared of successive differences in heartbeat intervals (RMSSD), low-frequency component, high-frequency component and their ratio (LF, HF, and LF/HF).In addition, we created a machine learning model capable of distinguishing between the stimulated and nonstimulated conditions from the ECG-derive data from various subjects and states. The model consisted of a deep convolutional neural network, which was trained on R-R interval (RRI) data extracted from ECG and time traces of LF, HF, LF/HF, SDNN, and RMSSD.Results:Only LF/HF ratio demonstrated a statistically significant change in response to stimulation. Although the LF/HF ratio is expected to increase during exposure to stress, we have observed that stimulation during exposure to stress counteracts this increase or even reduces the LF/HF ratio. This could be an indication that the vagus nerve stimulation decreases the sympathetic activation during stress inducement.Our Machine Learning model achieved an accuracy of 70% with no significant variations across the three states (baseline, stress, and recovery). However, training an analogous neural network to identify the states (baseline, stress, and recovery) proved to be unsuccessful.Conclusion:Overall, in this study, we showed further evidence of the beneficial effect of taVNS on stress processing. Importantly we have also demonstrated the promising potential of ECG metrics as a biomarker for the development of closed-loop stimulation systems.